If you want to participate, please vote for date and time for the first workshop

Please inform others in you community that will be interested in contributing to the success of this standard.

You will find more information on the workshops at http://wiki.teria.no/confluence/display/mlr5/Developing+MLR+Part+5+Educational

Technorati Tags: Dublin Core, metadata, MLR

First I would like to introduce the water ring methodology of writing this standard. The innermost ring represents the top level concepts. The rings further out are refinements of these concepts. Where you stop refining – at the third or fourth ring from the centre – is a matter of taste. If you keep going, you will be doing the work of the local communities that should create the application profiles. For the Part 5 of the MLR standards that is a few steps too far. It will be a challenge to get this message across, as every community (or National Body) wishes to have their pet elements represented in the “core” standard.

Speaking of which, I wonder how we should get consensus about postponing the Part on Core elements till we have a clearer picture of how the other parts will look like, e.g., the Educational, Technical, Rights, etc. Looking at the proposed top level concepts of the Educational part, there are some of these that would be common for several of the other parts. These concepts will have different refinements in the other parts, but they are still the same concepts, e.g. the Subject and the Agent, may be also the Audience.

In the diagram I have added some green boxes for the second (or third) ring of concepts. I am still not quite sure about how the relationship between the Audience and Educational Context will play out. Will it be crystal clear in which of these boxes you will be putting the refinements you come up with? As of now, it seems that Audience has a more “stable” characteristic, being nearer to what we could call authoritative metadata, i.e., metadata from the content creator’s perspective. The Context is more fluent, more context dependent (sic!), more susceptible to change – of contexts of use… Well, given these different characteristics of Audience and Educational Context, you might end up with the same refinements being part of both classes then?

Anyway, I have indicated that many of the LOM Educational elements, like Interactivity level and Typical Learning Time are part of the Educational Context class. If some of these are part of Instructional Model, I have not thought through. This way of reasoning shows how I think we should have discuss these matters in the month ahead - before we settle on the second working draft of the MLR Part 5 editorial group.

And lastly, the definitions. It goes without saying that we should reuse definitions from existing standards, i.e., the Dublin Core for Audience and Subject; IEEE LOM for some of the concepts inspired from that standard, and so on…

Technorati Tags: metadata, MLR, SC36

What is the current context of this standard? We have seen characteristics as “Balkanisations”, “proliferation of application profiles”, “great uncertainty as to the role of Content Repositories”, questions raised if open educational standards are still relevant when “content can be released in any format and hosted by multiple applications”, and so on.

The conclusions I draw from this, is that we should be very careful not to base our standardisation work only on a few use cases, mostly from the repository field of practice. We simply don’t know at the moment the different contexts of use that will benefit from learning resource metadata. Therefore, we need a scope for the Educational Part that is open to very flexible use of metadata.

Revisiting our first scope proposal from Working Draft 1, I think we kind of hit the nail. It is minimal and seems rather robust.

This Part 5 of the Metadata for Learning Resources Standard describes elements for tagging a learning resource to give the users a means to learn more about the pedagogical aspects of using the resource.

Moving out of the repository age to a situation where learning resources are scattered all over the web, we really don’t know how the metadata fragments are to be used. The Dublin Core community talks about “metadata islands” – with the vision that the semantic web will be able to interconnect these floating reeds of metadata fragments. Therefore, we should not bound it to too complex data models, but rather make sure that the semantic meaning of the fragments themselves are carried across applications and contexts.

The scope of the Part 5 should make it clear that

• use of metadata for education is about mixing and matching different metadata schemes - and that this standards is made with that in mind
• the use of educational metadata is an emerging field – a fact that should make us a little reluctant to draw up too elaborate models
• we should look for a “modular” or “thin” part

The previous scope proposal was rather verbose and carried all the excesses of the MLR work up till now:

Within the context of MLR Part 1 – Framework Model – this Part 5 provides information in the form of metadata elements about conditions pertaining to the pedagogical context, and other relevant aspects with respect to a learning resource. The focus here is on the provision of added description, i.e., via metadata elements (and their attributes), pertaining to the educational use and re-use, modification and/or distribution of a learning resource. Part 5 will indicate the educational or pedagogical attributes about the learning resources. Educational or pedagogic features of learning resources can change depending on the context in which that resources are used. Part 5 complements the core educational metadata elements of Part 2.

The question for the MLR Educational team is whether the development of the Part 5 should be based on the requirements from particular implementation uses cases (see the Dublin Core Application Profile work)? I would have been nice to have a few good use cases to support the development, as the consensus of what MLR is all about is not very stable at the moment.

The Dublin Core work gathers requirement from both Implementation Use Cases, and Functional Requirements. The latter seems to be of the same nature as MLR principles or rules from Part 1, but more clearly cut.

Technorati Tags: IEC JTC1 SC36 , metadata, MLR, SC36

At the present stage a study of JISC-CETIS’s Phil Baker has been of great help. The Learning Material Application Profile Scoping Study is available in draft from Phil’s blog. My proposal for the next editors meeting is to use the model Phil comes up with a starting point to discuss the top level concepts of MLR Educational.

First, some insights found in Phil’s report. He points out that metadata for education is one of the domains where the issues are least well articulated and where solutions are least developed. His study shows that where resource descriptions were seen as “difficult” there seems to be suggestions to move away from structured metadata towards the approach of providing semi-structured free text descriptions. This should be seen as a warning signal to SC36 colleagues who want to be too explicit in modelling educational metadata. We are making an international standard that should cover the needs of all educational levels in all regions and cultures throughout the world, a truly daunting task!

Phil is of the opinion that any future work should recognise that metadata may be valuable in supporting resource management, discovery/retrieval and use. And he reminds us that the application profile to support resource discovery might be quite different from an application profile to support resource management or use.

Therefore, we should come up with use cases that show a number of scenarios:

• The Provider role
• The Management role
• The Consumer role

The last role should not just cover the users of repositories in a strict sense, but also reflect the emergent use of learning resources through participation in different Communities of Practice, in digital living and serendipitous learning, i.e., the discovery use case.

Furthermore, Phil has a concept of the Primary Resource, the Learning Material; and the Secondary Entities that are related to the learning material and which have properties that need to be specified in order to provide full description of the primary resource. This is similar to my concepts of intrinsic and extrinsic properties of a learning resource. However, Phil uses the FRBR model to unpack the intrinsic qualities of the learning resource.

I have copied the model Phil Baker has made to illustrate what is described by metadata for learning materials. Whether we in MLR Part 5 should bring into play the full Functional Requirement for Bibliographic Records (FRBR) entity model is to be discussed. I would think that this will vary according to the needs of different communities, and that in some application profiles you will find the full model, and in others a reduced set.

What we should discuss, is this simple model, defining a learning resource by Subject, Audience, Agent and Educational Context.

We should list the metadata elements that the different National Bodies have launched as candidates for the new standard, and see if we are able to place them as refinements of these top level concepts, e.g., audience language, educational level, accessibility preferences, pedagogical modality, competency or learning outcome, etc.

The tricky bit will be to tackle the redundancies in this model, e.g., is educational level a property of Audience or Educational Context. Or is Audience in fact a property of Educational Context? I bet this discussion will show that this depends on the use scenarios, and that we will not be able to come up with a model that is completely streamlined at this stage of learning technology development.

Technorati Tags: metadata, MLR, SC36

Technorati Tags: metadata, Dublin Core

With the help of the king of abstract modelling, Ambjörn Naeve, we came up with this proposal for an abstract model:

You should open this model in Conzilla to explore the metadata of each concept.

The role of an application profile in this metadata approach is different from what we usually deal with in the IEEE LOM community. The proposed MLR Application Profile is more tied to the local context and communities, due to the fact that MLR is intended to be a true international standard, and the very concept of global interoperability is a oxymoron.

Therefore we modelled the MLR Application Profile as in the figure below.

This model should be opened and digested in Conzilla as well.

Peter Karlberg and Erlend Øverby looking at…

Ambjörn’s drawings with Tore Hoel as an observer

Technorati Tags: metadata, MLR

I bet the integer versus the real number problem is important when your doing eBusiness big time (even I don’t really see the point here either), but searching for learning resources? No way! It seems like the project is being hijacked by the eBusiness community and the learners and educators sidetracked. We should look more to the library community; ebusiness has very little to do with it! At least the library community is grappling with the same problems of going from the second order of things to the third, where everything is miscellaneous and what matters is not what fits on a library record card any more.

To me, the current MLR thinking has not yet reached the digital age. We are still thinking from the perspective of the content owner, who wants to control her resources, to allow the users to search, evaluate, retrieve and re-use the learning objects. But it is not about control. It is not about filtering what the users should be allowed to see. Filtering on the way out, not on the way in, is a new strategic principle of the third order of things, according to David Weinberger in his Everything is miscellaneous (p. 102). He has three more principles, one already mentioned: Give up control!

The two other principles, Put each leaf on as many branches as possible; and Everything is metadata and everything can be a label, addressed the very model of the MLR. We should not create a huge, complicated structure bound to a XML tree, trying to capture the essence of the physical learning resource itself. Instead we should be very aware of identifiers, and then try to capture as many small bits of characteristics as possible, storing them in as many places and ways a possible. Living with learning resources is not a question of exactness (in the eBusiness manner), but in finding the just right level of fuzziness (in the learning for life manner of using resources). We then need a system that identifies what we are talking about, and leaves the other aspects up to the users to negotiate. If a community wants to be very specific in their negotiation, they should be allowed to do so. However, if others are coming by, the conversation should be digitally captured in a way that degrades gracefully. If my system cannot make sense of the value of the element “semantic density”, it should at least let the user know that the discussion is about Educational Context. The choice of tags should be left to the user communities. The constraining technology should be as general as possible; however, based on principles that allow computers to reason and suggest combinations and results to the users.

It is time for the MLR work to realize that computers are great for finding and combining, if they have anything to work on. Providing metadata is just about this: Spreading leaves of information around, to give the computers stuff to work on. Not exact to the 10th decimal; it is enough that it is about something, and is identifiable.

Technorati Tags: metadata, ISO/IEC JTC1 SC36

Now, reading the committee draft outlining the second part on “core elements” I realize that the thinking of the workgroup is too narrow - and the user requirements perhaps not reflecting today’s practice - or should we say, web architecture. The scope is now stated as “directed at maximizing interoperability of learning resource records (MLRR) at the time of federated search and metadata harvesting”. Hey, is not the scope of metadata wider than repository management and institutional interests? It seems that the perspective from the use of metadata for learning and the use of metadata needed to fulfill the promises of the semantic web and web architecture is completely lost!

Technorati Tags: metadata, web standards

This presentation addresses the repository situation in Norway well, in the respect that we are only discussing open access and focussing on scholarly publications - at least in HE. Andy is explaining why this perspective is too narrow - and what we should look for….

Something else: My apologies for not being an active blogger lately. I promise to change, as much as this autumn will be a very active one - working on the upcoming eContentplus Best Practice Network Icoper. More information coming, but I am now introducing a new blog category - icoper. Stay tuned!

Well, perhaps that is what Scott Wilson is referring to when he is commenting Wayne Hogdgins presentation of the Future of Metadata. Scott says “There is some good stuff in here on future work, but to be honest it would be better off ditching most of the legacy rather than trying to convince us of what a wonderful success its been, and all this new stuff is logical progression.”

It is a growing consensus among the “e-learning futurists” that we got learning objects and metadata wrong at first try. That seems to dawn upon some of the founding fathers of LOM and some of the other standards now. It is just interesting to see how this is manifested in their presentations around the world.